The Translation Hub transforms governance established within human society—including laws, regulations, guidelines, standards, and best practices—into forms that AI systems can use. The problem in AI governance is not necessarily the absence of rules. In many cases, the relevant rules already exist. However, they are expressed in forms intended for human interpretation and implementation, including legal texts, administrative documents, technical standards, contracts, and internal organizational rules.

Moreover, there is no single body of governance that applies universally. As illustrated by Country A, Country B, and Country C, laws and regulations differ across the countries and regions in which AI is used. Even the same AI Provider must therefore operate across multiple institutional environments. Applicable conditions also vary according to industry, use case, organization, and user. Consequently, embedding individual rules directly and permanently within an AI model makes it difficult to accommodate continuous institutional change or deployment across multiple environments.

The Translation Hub is positioned between the governance of human society and AI systems. It analyzes institutional requirements expressed in Human Language and structures them into machine-readable forms that AI can process. The resulting information is represented as Policy Tags, which can provide common governance information to different models, services, agents, and use cases operated by an AI Provider.

Importantly, the Translation Hub is not a mechanism for “determining” laws or policies themselves. The Authority that establishes an institution is separated from the function that translates its requirements into a form applicable to AI systems. Accordingly, when an institution changes, the change can be reflected by updating the corresponding governance information, without redesigning the AI model itself.

The essential function depicted here is therefore not merely the conversion of Human Language → Machine-readable.

Institution → Translation → Machine-readable Governance → AI Systems

It is the establishment of this connective pathway. This makes it possible to govern numerous AI systems operating across different countries, institutional environments, and use cases without confining governance within individual models.